# Run before lecture to load datasets and do simple prep
library(tidyverse) #all our data wrangling/plotting
options(repr.matrix.max.rows = 6)
# Making scatter points a bit bigger so that students can see them
update_geom_defaults("point", list(size = 3))

#Mauna Loa
co2_df <- tibble(
    concentration = as.vector(co2),
    date = lubridate::date_decimal(as.numeric(time(co2)))
)

#Top 12 Island landmasses
islands_df <- enframe(islands)
colnames(islands_df) <- c('landmass', 'size')
islands_df = top_n(islands_df, 12, size)

continents <- c('Africa', 'Antarctica', 'Asia', 'Australia', 'Europe', 'North America', 'South America')
islands_df <- mutate(islands_df, is_continent = ifelse(landmass %in% continents, 'Continent', 'Other'))

gapminder <- read_csv("data/gapminder.csv")
gapminder_2016 <- gapminder  |>
    select(country, year, continent, life_expectancy) |>
    filter(year == 2016)

#old faithful, mtcars -- nothing to do

Warning message:
“package ‘ggplot2’ was built under R version 4.3.2”
── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.3     ✔ readr     2.1.4
✔ forcats   1.0.0     ✔ stringr   1.5.0
✔ ggplot2   3.5.0     ✔ tibble    3.2.1
✔ lubridate 1.9.2     ✔ tidyr     1.3.0
✔ purrr     1.0.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
Rows: 10545 Columns: 9
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (3): country, continent, region
dbl (6): year, infant_mortality, life_expectancy, fertility, population, gdp

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

DSCI 100 - Introduction to Data Science

Lecture 4 - Data visualization in R

Attribution: images in these slides that are not accompanied by code mostly come from
The Fundamentals of Data Visualization by Claus O. Wilke

Artwork by @allison_horst

Today: Visualization

image source: R for Data Science by Grolemund & Wickham

Designing a visualization: ask a question, then answer it

The purpose of a visualization is to answer a question about a dataset of interest.

A good visualization answers the question clearly. A great visualization also hints at the question itself.

Visualizations alone help us answer two types of questions:

• descriptive: What are the largest 7 landmasses on Earth?
• exploratory: Is there a relationship between penguin body mass and bill length?
• inferential
• predictive
• causal
• mechanistic

(we need more tools + visualizations to answer the others)

• Descriptive: A question which asks about summarized characteristics of a data set without interpretation (i.e., report a fact). (describe characteristics)

• Exploratory: A question asks if there are patterns, trends, or relationships within a single data set. Often used to propose hypotheses for future study. (discovery of ideas and thoughts)

• inferential: determine if association observed in your exploratory analysis hold in a different sample that is rep of pop (infew what is true)

• predictive: what predicts whether someone will eat a certain diet

• causal: whether changing one factor will change another factor

• mechanistic: how e.g. how diet leads to a reduction in the number of viral illnesses

Creating visualizations in R

• It's an iterative procedure. Try things, make mistakes, and refine!

• We will use ggplot2. There are three key aspects of plots in ggplot2:

  1. aesthetic mappings: map dataframe columns to visual properties
  2. geometric objects: encode how to display those visual properties
  3. scales: transform variables, set limits

• Add these one by one using +

Types of variables

A variable refers to a characteristic of interest and can be:

1. categorical: can be divided into groups (categories) e.g. marital status
2. quantitative: measured on a numeric scale (usually units are attached) e.g. height

Scatter Plots

To visualize the relationship between two quantitative variables

e.g. Is there a relationship between horsepower and fuel economy of an engine? Does the number of cylinders affect that relationship?

# Load libraries for wrangling and plotting
library(tidyverse)

# Inspect the data
mtcars

A data.frame: 32 × 11

	mpg	cyl	disp	hp	drat	wt	qsec	vs	am	gear	carb
	<dbl>	<dbl>	<dbl>	<dbl>	<dbl>	<dbl>	<dbl>	<dbl>	<dbl>	<dbl>	<dbl>
Mazda RX4	21.0	6	160	110	3.90	2.620	16.46	0	1	4	4
Mazda RX4 Wag	21.0	6	160	110	3.90	2.875	17.02	0	1	4	4
Datsun 710	22.8	4	108	93	3.85	2.320	18.61	1	1	4	1
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮
Ferrari Dino	19.7	6	145	175	3.62	2.77	15.5	0	1	5	6
Maserati Bora	15.0	8	301	335	3.54	3.57	14.6	0	1	5	8
Volvo 142E	21.4	4	121	109	4.11	2.78	18.6	1	1	4	2

# Set the default size for all plots
options(repr.plot.width = 10, repr.plot.height = 8)

# Is there a relationship between fuel economy and horsepower?

Build up one-by-one:

# Base 
ggplot(mtcars, aes(x = hp, y = mpg)) +
    geom_point()
    
# It's difficult to see the text so let's make it bigger 
theme(text = element_text(size = 26))

As horsepower increases miles per gallon (fuel efficiency) tends to decrease (negative relationship). But is this true for all cars? Can we group the data in some way to find out more? What about per the number of cylinders (the size) of the engine?

# Color per cylinder
ggplot(mtcars, aes(x = hp, y = mpg, color = cyl))

# Change color to factor/categorical
ggplot(mtcars, aes(x = hp, y = mpg, color = factor(cyl)))

Cars with more cylinders tend to have higher horsepower and lower fuel efficiency. We can make this plot easier to understand by adding axis labels:

# Add labels
labs(x='Horsepower', y='Miles per Gallon', color='Cylinders')

# Full final plot
ggplot(mtcars, aes(x = hp, y = mpg, color = factor(cyl))) +
    geom_point() + 
    theme(text = element_text(size = 26)) +
    labs(x='Horsepower', y='Miles per Gallon', color='Cylinders')

Line Plots

To visualize trends with respect to an independent quantity

e.g. How has atmospheric carbon dioxide changed over the last 40 years?

Mauna Loa Research Station

# Inspect the data
co2_df

A tibble: 468 × 2

concentration	date
<dbl>	<dttm>
315.42	1959-01-01 00:00:00
316.31	1959-01-31 10:00:00
316.50	1959-03-02 20:00:00
⋮	⋮
360.83	1997-10-01 18:00:00
362.49	1997-11-01 04:00:00
364.34	1997-12-01 14:00:00

dttm = Datetime column type

# Change the default text size for all plots
theme_set(theme_gray(base_size = 26))

# How does atmospheric CO2 concentration change over time?

Start with geom_point as we just learned:

ggplot(co2_df, aes(x=date, y=concentration)) +
    geom_point()

The visualization shows a clear upward trend in the atmospheric concentration of CO2 over time. However, something is not quite right here; ask students what wrong with this plot (overplotting).

Switch to geom_line:

ggplot(co2_df, aes(x=date, y=concentration)) +
    geom_line()

Share additional conclusion: The concentration seems to oscillate as well.

Optionally add labels or just mention it:

    labs(x = 'Date', y = 'CO2 Concentration')

Bar Plots

To visualize the comparison of amounts

e.g. Which are the largest 12 island landmasses on Earth? Are they all continents or are there some other islands with large landmasses as well?

Source: Wiktionary

# Inspect the data
islands_df

A tibble: 12 × 3

landmass	size	is_continent
<chr>	<dbl>	<chr>
Africa	11506	Continent
Antarctica	5500	Continent
Asia	16988	Continent
⋮	⋮	⋮
New Guinea	306	Other
North America	9390	Continent
South America	6795	Continent

# What are the largest 12 island landmasses on Earth?

Simplest approach first:

ggplot(islands_df, aes(x = landmass, y = size)) +
    geom_bar()

Challenging question: Think back to the pre-reading, why do we get an error message? Because geom_bar is only for counts by default:

ggplot(islands_df, aes(x = is_continent)) +
    geom_bar()

To make geom_bar use the y aesthetic, we need to change the stat option to identify (use the value as is, don't count anything).

ggplot(islands_df, aes(x = landmass, y = size)) +
    geom_bar(stat = 'identity')

The x-axis labels are overlapping. It is possible to rotate them, but that would make them hard to read. A more effective approach is to change so that the labels are on the y-axis.

ggplot(islands_df, aes(x = size, y = landmass)) +
    geom_bar(stat = 'identity')

To make the plot easier to read, we can reorder the bars by size. Generally we put the largest bar closest to the axis, but this is not a hard rule.

ggplot(islands_df, aes(x = size, y = reorder(landmass, -size))) +
    geom_bar(stat = 'identity')

Mention that we should change the labels to meaningful names and optionally show it. Optionally color bars by continent as the last step.

Histograms

To visualize the distribution of a single quantitative variable

e.g. Was there a difference in life expectancy across different continents in 2016?

# Inspect the data
gapminder_2016

A tibble: 185 × 4

country	year	continent	life_expectancy
<chr>	<dbl>	<chr>	<dbl>
Albania	2016	Europe	78.1
Algeria	2016	Africa	76.5
Angola	2016	Africa	60.0
⋮	⋮	⋮	⋮
Yemen	2016	Asia	64.92
Zambia	2016	Africa	57.10
Zimbabwe	2016	Africa	61.69

# Was there a difference in life expectancy across different continents in 2016?

First let's create a histogram of all countries' life expectancies.

ggplot(gapminder_2016, aes(x = life_expectancy)) +
    geom_histogram()

Then color by continent. Note that we need to use fill since color is just for the outline of each rectangle.

ggplot(gapminder_2016, aes(x = life_expectancy, fill = continent)) +
    geom_histogram()

The position of the bars defaults to "stack" that they are stacked on top of each other - not very easy to read. Setting the position to identity will overlay the histograms on top of each other so that they all have the same baseline. When doing this we also need to add transparency.

ggplot(gapminder_2016, aes(x = life_expectancy, fill = continent)) +
    geom_histogram(position = "identity", alpha = 0.6)

This is still not an effective way to convey this information, so let’s try a different strategy of creating multiple separate histograms instead using facetting.

ggplot(gapminder_2016, aes(x = life_expectancy, fill = continent)) +
    geom_histogram() +
    facet_grid(rows = vars(continent))

That looks much better and is easy to compare! (Mention that axis labels should be changed to be human readable and optionally show it as well as how to make the plot taller)

options(repr.plot.width = 10, repr.plot.height = 12)
...
    labs(x = "Life Expectancy (years)", y = "Count", fill = "Continent")

A few rules of thumb for creating effective visualizations

Rule of Thumb: No tables / pie charts / 3D

Which one is easier to interpret? Pie graph - colours don't mean anything (unneccessary)

• hard to see size of slices relative to the other slices

Rule of Thumb: No tables / pie charts / 3D

• the third dimension does not improve the reading of the data
• these plots are difficult to interpret because of the distorted effect of perspective associated with the third dimension.
• 3D is discouraged for charts in general, and should only be used for very specific applications
• the bars or slices in a pie graph that are closer to the reader appear to be larger than those in the back due to the angle at which they're presented

Rule of Thumb: Use simple, colourblind-friendly colour palettes

Rule of Thumb: Include labels and legends, make them legible

Remember: a great visualization tells its own story without needing you to be there explaining things

options(repr.plot.width = 4, repr.plot.height = 4)
diamond_plot <- ggplot(diamonds, aes(x = carat, y = price)) +
    geom_point() +
    xlab("Size (carat)") +
    ylab("Price (US dollars)")
diamond_plot

Add alpha = 0.2 to geom_point

• transparency setting between [0,1]

• too many colours (overwhelming)

• less is more

• Make sure to use colourschemes that are understandable by those with colourblindness. For example, the RColorBrewer R library provide the ability to pick such colourschemes, and you can check your visualizations after you have created them by uploading to online tools such as the colour blindness simulator.

• Redundancy can be helpful; sometimes conveying the same message in multiple ways reinforces it for the audience. For instance you can also consider using shapes to represent different groups

Go and create!

What did we learn today?

TidyTuesday!

Weekly practice exploring and visualizing data: https://github.com/rfordatascience/tidytuesday